jialicheng/unlearn_cifar100_swin-base_random_label_6_87

87

This model is a fine-tuned version of microsoft/swin-base-patch4-window7-224 on the cifar100 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4653
• Accuracy: 0.8882
• Dt Accuracy: 0.8882
• Df Accuracy: 0.0283
• Unlearn Overall Accuracy: 0.9697
• Unlearn Time: None

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0002
• train_batch_size: 128
• eval_batch_size: 256
• seed: 87
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Overall Accuracy	Unlearn Overall Accuracy	Time
No log	1.0	391	0.4979	0.8923	0.2091	0.2091	None
1.3347	2.0	782	0.5263	0.8753	0.2354	0.2354	None
1.1228	3.0	1173	0.4918	0.865	0.2513	0.2513	None
1.02	4.0	1564	0.4820	0.8643	0.2522	0.2522	None
1.02	5.0	1955	0.5010	0.7847	0.3634	0.3634	None
0.9171	6.0	2346	0.4821	0.7063	0.4599	0.4599	None
0.8197	7.0	2737	0.4831	0.6257	0.5466	0.5466	None
0.7522	8.0	3128	0.4938	0.5033	0.6597	0.6597	None
0.6776	9.0	3519	0.4919	0.4307	0.7173	0.7173	None
0.6776	10.0	3910	0.4964	0.3443	0.7819	0.7819	None
0.6029	11.0	4301	0.5227	0.279	0.8225	0.8225	None
0.5621	12.0	4692	0.5162	0.22	0.8584	0.8584	None
0.5102	13.0	5083	0.4961	0.1737	0.8885	0.8885	None
0.5102	14.0	5474	0.4890	0.1083	0.9225	0.9225	None
0.4579	15.0	5865	0.4872	0.0813	0.9398	0.9398	None
0.4165	16.0	6256	0.4732	0.0703	0.9468	0.9468	None
0.3853	17.0	6647	0.4681	0.0587	0.9530	0.9530	None
0.3564	18.0	7038	0.4745	0.0347	0.9633	0.9633	None
0.3564	19.0	7429	0.4679	0.0327	0.9665	0.9665	None
0.3261	20.0	7820	0.4653	0.0283	0.9697	0.9697	None

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.2+cu118
• Datasets 2.18.0
• Tokenizers 0.15.2